I have been doing bug hunting almost all day, two bug fixes are enclosed,
for the remainder I am willing to stick my head into the code next week
until the performance is up to a reasonable level and most of the known
bugs are ironed out.
I went through the earlier patches and tossed out some material and re-arranged
the TFRC lookup stuff in a different way. I think it is now much easier to read.
Patch 1: Bug fix: sending rates were calculated before X_recv/p were updated
Patch 2: Removes the `Illegal ACK received message' [it is harmless]
Patches 3..8 correspond to the earlier TFRC patches (much smaller now):
Ian I would appreciate if you could have a look at these.
I have put in effort to explain them clearer. While the performance still
is bad, these are a step forward. In particular, the parameter checking is
useful to have. Should it be needed, I have documented all everything on
http://www.erg.abdn.ac.uk/users/gerrit/dccp/tfrc_documentation/doc/
Patch 3: just adds documentation for the TFRC lookup table; no code change, but
this provides the basis for all other patches (the formulas)
Patch 4: corrects a small error in the reverse lookup function; I
have now added a detailed explanation _why_ this is an error.
Patch 5: adds extensive parameter checking to the TFRC lookup routines.
Patch 6: wraps up the previous steps by simplifying the forward lookup.
It also defines a constant for the lowest possible resolution and
adds a warning if p is greater than 0 and less than this threshold.
Patch 7: deprecates the use of TFRC_SMALLEST_P in ccid3.c. It is subsumed
by the minimum threshold set in Patch 4. The use of TFRC_SMALLEST_P
is dangerous, as it masks large errors for low values of p.
The resolution error is exactly given by:
(f(0.0001) - f(p))/f(p) * 100
and for TFRC_SMALLEST_P = 0.00004 this is still a value of about 58%!
It is better to observe the warning messages and generate a new table
in case there are too many warnings.
Patch 8: adds bin-search which should speed up TFRC reverse-lookup a little
-
To unsubscribe from this list: send the line "unsubscribe dccp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
